A 10-Minute Audit: Does ChatGPT, Gemini or Perplexity Mention Your Business?

This is a checklist you can genuinely run in the next ten minutes, using nothing but a phone and the free tier of a few AI assistants. It won’t fix anything by itself, but it will tell you, honestly, whether a customer typing a natural question into ChatGPT, Gemini or Perplexity would ever hear about your business — which is worth knowing before spending money on anything to improve it. Share the result with a colleague or business partner once you’ve run it; a second person testing the same questions independently often catches a mention or a gap the first pass missed, since AI assistants don’t always give the exact same answer twice.

Before you start: what you need

No sign-up or paid tool is required. Open the free (or logged-out) tier of ChatGPT, Gemini and Perplexity in separate tabs on a phone or laptop, and have two things ready: the plain-language category your business falls under (not your brand name — "aircon servicing", not "ABC Aircon Sdn Bhd"), and the specific service or product you’re actually known for. Knowing your two or three closest named competitors also helps, since seeing whether they appear instead of you is often more informative than an empty answer.

Step 1: ask each assistant the way a real customer would

Open ChatGPT, Gemini and Perplexity, one at a time, and ask two things in each:

• “Who is a good [your business type] in [your city]?” — e.g. “Who is a good web design agency in Petaling Jaya?”
• “Recommend a [your service] in [your area].” — e.g. “Recommend a dental clinic in Subang Jaya.”

Use plain, natural phrasing rather than your exact business name — the point is to see what comes up for someone who doesn’t already know you exist. Try a second, more specific version too, using whatever product or service the business is actually known for, since a broad question and a specific one can surface different results.

Step 2: record whether you’re mentioned, honestly

Keep the record simple — a business type, an assistant, and one of three outcomes is all that’s needed, and it takes seconds per attempt once the pattern is set.

For each assistant and each question, note one of three outcomes: mentioned by name, mentioned only if you ask a follow-up naming your business directly, or not mentioned at all even after a follow-up. This is the part worth being honest about — it’s tempting to count a mention that only appears after leading the AI straight to your name, but that isn’t the same as a customer discovering you unprompted.

Step 3: if you're not mentioned, work out why

A handful of causes account for most cases of a real, established business simply not coming up:

• Thin or vague website content. A homepage that says what a business is proud of but never plainly states what it does, where, and for whom gives an AI assistant little concrete to extract.
• No structured content or schema markup. AI tools extract facts more confidently from clearly labelled information — services, service areas, FAQs — than from unstructured marketing prose.
• Inconsistent NAP (Name, Address, Phone) details. If the business name, address or phone number differs even slightly across the website, Google Business Profile and directory listings, an AI assistant has a harder time confirming which version is correct.
• No FAQ blocks. A written FAQ section hands an assistant a ready-made question-and-answer pair it can lift almost directly into a response.
• Few or generic reviews. Reviews that describe specifics — what was done, how it went — carry more weight as a signal than a high star rating with little detail behind it.

Why the three assistants can give different answers

It’s worth running the audit across all three rather than just one, because ChatGPT, Gemini and Perplexity don’t all find businesses the same way — some lean more on live web search, others draw more on a fixed body of training data or a specific data source like Google or Bing. Perplexity is built around live web search with visible citations, so it tends to reflect whatever is currently indexed and well-structured; Gemini can draw on Google's own search and Maps data, which is part of why consistent Google Business Profile details matter for it specifically; ChatGPT blends training knowledge with live browsing depending on the query, so its answer can shift between a well-known chain and a niche local business in ways the other two don’t always match. A business missing from one assistant’s answer but present in another isn’t a contradiction; it’s a sign of which underlying sources each tool is actually pulling from, and it means a single “we’re not on AI” conclusion from testing only one assistant can be misleading in either direction.

A worked example

Say a bakery in Ipoh asks Perplexity “Recommend a cake shop in Ipoh” and gets three named competitors back, none of them the bakery itself, each with a short reason attached — a specialty, a location, a standout product. That specificity is the tell: those three businesses likely have clear, structured facts an AI could confidently extract, while the missing bakery either doesn’t, or has facts that are too vague or inconsistent to state with confidence. Running the same question in ChatGPT and Gemini and seeing whether the pattern repeats is what turns one data point into an actual diagnosis rather than a fluke.

A second example shows the opposite pattern: a Klang Valley clinic asks all three the same two questions and is named by Perplexity and Gemini but not ChatGPT. Because the clinic has a well-maintained Google Business Profile with consistent NAP details and a steady stream of detailed reviews, the Google-adjacent tools surface it confidently; the gap in ChatGPT is worth investigating separately rather than assumed to be the same root cause — perhaps the website itself lacks the structured service pages ChatGPT's browsing would otherwise pick up.

What a “good” result actually looks like

Being named in every single answer to every phrasing isn’t a realistic bar, even for an established business — these tools vary their answers, and there’s rarely only one correct recommendation for a search like this. A genuinely good result is being mentioned unprompted in most attempts across at least two of the three assistants, with the facts stated about the business being accurate. Being mentioned in one out of six attempts, or only after directly naming the business, points to a real visibility gap worth addressing rather than a fluke of how the question was worded.

Running both questions in all three assistants gives six attempts total — a simple way to turn that into a score:

A trap worth avoiding

Once a gap shows up, it’s tempting to chase quick wins that look like they’d help — stuffing a homepage with every possible keyword variation, or buying a batch of generic five-star reviews. Both tend to backfire. Keyword-stuffed copy reads as less clear and specific to an AI tool trying to extract a genuine fact, not more; and reviews that are vague or clearly bulk-purchased carry little of the specific detail that actually functions as a signal, sometimes flagged outright by the platforms hosting them. The fixes that actually work are the boring, structural ones — clear facts, consistent details, real detailed reviews — not a shortcut around them.

What to actually do with the results

If the audit turns up a gap, the fixes map fairly directly to the causes above: tighten the website’s copy so it plainly states what the business does and for whom, add schema markup and structured content an AI tool can extract with confidence, check that the business’s name, address and phone number match exactly everywhere they’re listed, add a proper FAQ section, and encourage specific, detailed reviews rather than just a star rating. None of these are one-off fixes — re-running this same ten-minute audit every few months is the honest way to see whether they’re actually working.

For more detail on how these tools decide what to surface in the first place, see our guide on how AI search engines pick which businesses to recommend, which goes deeper into each of the causes above, including how ChatGPT, Perplexity, Gemini and Copilot each source information. None of these fixes require touching all five causes at once — tackling NAP consistency and a proper FAQ section first is usually the fastest route to a measurable change, since both are concrete, one-off corrections rather than an ongoing content commitment.

Common mistakes

• Testing only one assistant. ChatGPT, Gemini and Perplexity draw on different sources; a gap in one isn't proof of a gap everywhere.
• Leading the AI straight to your own name. A mention that only appears after asking about the business directly isn't the same as an unprompted recommendation.
• Running the audit once and never again. AI assistants update constantly; a result from months ago may no longer be accurate.
• Keyword-stuffing the homepage as a quick fix. It reads as less clear to an AI tool trying to extract a genuine fact, not more.
• Buying generic reviews to fill the gap. Vague or bulk-purchased reviews carry little real signal and are sometimes flagged by the platforms hosting them.
• Treating one low score as proof nothing can be done. Most gaps trace back to a short, fixable list: thin copy, missing schema, inconsistent NAP, no FAQ, generic reviews.

Where Gotka Technologies fits

Gotka’s web design packages include Generative Engine Optimisation as an add-on — GEO Basic at RM1,200 or GEO Advanced at RM4,200 — structured content and schema markup built specifically to help AI search tools extract and state a business’s facts with confidence, which is exactly the gap this audit is designed to surface. It sits alongside the same basic SEO every Gotka website already includes at no extra charge, rather than replacing it. For businesses that also want an ongoing supply of new content for AI tools to cite, the SEO content retainer’s top tier, SEO Accelerate at RM3,200/month, includes full GEO content tuning as part of its ongoing posts.

Key terms used in this guide

• AI visibility: whether AI assistants such as ChatGPT, Gemini and Perplexity mention a business unprompted when answering a relevant question.
• GEO (generative engine optimisation): structuring and writing content so AI answer engines can extract and cite it confidently.
• NAP (Name, Address, Phone): the basic identifying facts about a business, which need to match exactly everywhere they're listed.
• Schema markup: structured code added to a webpage that labels facts, such as services or FAQs, so machines can read them reliably.
• Unprompted mention: being named by an AI assistant without first asking about the business by name.
• Structured content: information presented as clearly labelled facts — headings, lists, FAQs — rather than unstructured prose.
• Google Business Profile: a business's free Google listing, a common source AI tools and search draw facts from.
• Visibility gap: the difference between how often a business should reasonably come up in AI answers and how often it actually does.